I'm working with a board with the AD9963. In debugging our designs, I've observed some behaviors different from those described in the "Rev. 0" datasheet.
I believe AD9963's interpretation of the following register bits is inverted from the datasheet descriptions. I've observed the following behaviors in hardware:
- 0x31: TX_BNRY; setting this to 1 configures the DACs to interpret samples as two's complement
- 0x32: RX_BNRY; setting this to 1 configures the ADCs to output two's complement samples
- 0x51: RX_PTTRN; setting this to 1 drives the checkboard pattern to TRXD
- 0x75: DLL_RESB; if this bit is 1, the DLL never locks. If it's 0 (i.e. I treat the reset as active-high), the DLL works as expected.
I've observed odd I/Q DAC behavior when changing RSET_SEL (reg 0x62). When RSET_SEL=1, the full scale current of the I and Q DAC outputs rises very slowly, increasing by >1mA in ~3 minutes.
Our board has a 10k/1% resistor to ground at pin 6 (RXBIAS) and a 0.1uF X5R cap to ground at pin 63 (REFIO).
A few observations:
- If I leave RSET_SEL=1 for a few minutes and let the full scale current rise, then set RSET_SEL=0, the full scale current resets to its nominal level and remains constant.
- If I leave RSET_SEL=1 for a few minutes and let the full scale current rise, then set RSET_SEL=0, then set RSET_SEL=1, the full scale current briefly resets to its nominal level (while RSET_SEL=0), then starts rising again, starting over from its nominal level.
- I also measured the voltage at REFIO (pin 63) during these tests. When RSET_SEL=0, V_REFIO=1.0V; when RSET_SEL=1, V_REFIO=~20mV.
- I see the same behavior at REFIO when changing TXDAC_PD reg 0x60; when TXDAC_PD=1, REFIO=~20mV; when TXDAC_PD=0, REFIO = 1.0V.
- However when TXDAC_PD=1, the I/Q DAC full scale current drops to ~0.1mA (or something similarly low; this is hard to measure with my equipment).
Is the DAC voltage reference somehow dependent on RSET_SEL, in addition to TXDAC_PD?
- Fig. 70: the label for the N block says "M[3:0]"
- Pg. 36: the decimation filter stopband rejection is described as 7dB (should be 70dB?)